The core issue that makes managing a population’s health is that by asking providers to carry out population health management, we’re asking them to take on activities that are outside their traditional role, training and daily workflow. It requires checking in with patients outside of the clinical setting and addressing barriers such as financial issues, transportation issues, mental health problems, health literacy issues and abnormal readings from home devices. How can they do this when their clinic is already full of patients? Well, technology will be a big part of the answer, but as we saw at Acupera, it won’t be enough. However, the better the technology gets in off-loading more and more from the providers’ to-do list, the more likely it will be that providers take it up and successfully execute these programs.
What if a lot of the checking-in, numerical analysis and education could be done automatically? What if the data could be collected passively or with minimal effort? Algorithms could then analyze the data and figure out what needs to be done, address basic issues, create plans and assign activities to the appropriate resources.
This is where sensors, wearables, AI and telehealth come in. As these technologies mature and their use cases become better formulated, they can become the cornerstone of population health management. They can take on the extra work that the providers are unable to do at scale. This means that in the near future, we should be able to identify and manage health risk much better and at scale, improving people’s health at a fraction of what we’re spending today.
One of the reasons why AI-based digital health tools have the potential to be so valuable to us is that they have the potential to remove the access barriers that are stopping us from reaching more people. Better still, AI can help to reduce the day-to-day load of managing chronic conditions, both for the provider and the patient. In my opinion, the key reason for the mixed results of population health programs so far is the fact that ongoing care management requires many activities to be performed by the patient and the care team and there’s just not enough people or time to do it all. Providing intelligent automation that takes care of a lot of these tasks would go a long way towards ensuring the success of these programs.
This is one of the areas where I’m the most optimistic about generative AI (e.g. ChatGPT) in healthcare, because it could eventually power meaningful interactions with patients. If responding to patient questions can be handled by generative AI, that removes the burden of the care team having to respond to every patient issue. And if generative AI can create discharge summaries and make patient care plans more personalized, it will improve the odds of patients adhering to those plans. When a telephone encounter is created, the assigned nurse will be able to use generative AI to create a response based on the patient’s message. The tool will provide recommendations based on what the patient is calling for, the nurse will review the choices, select the response and add it to the record with a single click.